[R] optim

[Ken Beath]

The function needs to have a single parameter. Then extract each
parameter.
For example, or see first example in help for optim.
errorFunction <- function(params) {
   localShifts <- params[1]
etc

>>> "Gerster  Sarah" <sgerster at student.ethz.ch> 01/03/07 9:22 PM >>>
Hi!

I'm trying to figure out how to use optim... I get some really strange
results, so I guess I got something wrong.

I defined the following function which should be minimized:
errorFunction <-
function(localShifts,globalShift,fileName,experimentalPI,lambda)
{
  lambda <- 1/sqrt(147)
  # error <- abs(errHuber(localShifts,globalShift,
  #   
"/home/sarah/Semesterarbeit/Sequences/R/R1593_filtered.data",3.48)) +
  #    sum(abs(localShifts))*lambda

  error <- sum(abs(localShifts))*lambda
  error # return the error to be minimized
}

Then I call optim:
par <- seq(length=9, from=0, by=0)
lambda <- 1/sqrt(147)
optim(par, errorFunction, gr=NULL, method="Nelder-Mead", hessian=FALSE,
        globalShift,
        "/home/sarah/Semesterarbeit/Sequences/R/R1593_filtered.data",
        experimentalPI=3.48, lambda = lambda)

The output is:
$par
[1]  0.56350964  0.56350964  0.56350964  0.56350964  0.00000000
-0.29515957
[7]  0.00569937  0.32543297  0.18615880

$value
[1] 0.2529198

$counts
function gradient 
      31       31 

$convergence
[1] 0

$message
[1] "CONVERGENCE: REL_REDUCTION_OF_F <= FACTR*EPSMCH"

Warning messages:
1: bounds can only be used with method L-BFGS-B in: optim(par,
errorFunction, gr = NULL, method = "Nelder-Mead",  
2: NAs introduced by coercion 



If I change my "error-function" to
errorFunction <-
function(localShifts,globalShift,fileName,experimentalPI,lambda)
{
  error <- sum(abs(localShifts*lambda))
  error # return the error to be minimized
}

or to:
errorFunction <-
function(localShifts,globalShift,fileName,experimentalPI,lambda)
{
  error <- sum(abs(localShifts))/sqrt(147)
  error # return the error to be minimized
}

The output is:
$par
[1]  6.018101e-20  6.018101e-20  6.018101e-20  6.018101e-20 
0.000000e+00
[6]  5.176245e-21 -4.002183e-21 -8.254019e-20  3.231412e-21

$value
[1] 2.768593e-20

$counts
function gradient 
      76       76 

$convergence
[1] 0

$message
[1] "CONVERGENCE: NORM OF PROJECTED GRADIENT <= PGTOL"

Warning messages:
1: bounds can only be used with method L-BFGS-B in: optim(par,
errorFunction, gr = NULL, method = "Nelder-Mead",  
2: NAs introduced by coercion 


- What is wrong with the first version?

Thanks for the help!

Sarah

______________________________________________
R-help at stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.